The problem that I encountered in doing such a build from the Stereophile article is that not only is the monolithic buffer used there not readilly available, but the successor to it is also not available. Further, of the currently available buffer chips, all of them (like the Burr Brown buf634 or the Intersil HA3-5002) seem to have something about them that make them poor choices for this type of projetct -- prmarily that they just don't sound that good.

This leaves a couple of additional options: The first is to use an opamp that is stable at unity gain. Opamps, however, suffer from the same problems as existing monolithic buffers (that they sound bad). Further, the best opamps are not typically stable at unity gain anyway, and thus, this would become an active preamp. The two remaining options, however, are both good. These are to use a discrete solid state design, or to use a tube design like a cathode follower. For this project I will use a solid state design.

There are a number of common discrete buffer designs. The most common being the so-called diamond buffer. To keep the build as simple as possible, however, I chose to use a slightly simpler class A buffer. This design is said to be originally designed by Les Sage, and has been popularized by Nuuk at Decibel Dungeon as a preamp for Gainclone power amplifiers. For a thorough background, you should read all about the project there. This is an excellent resource, particularly if you are just beginning.

I have built a number of these and tried a number of different tweaks and parts changes. The first change I made was the inclusion of R1. This seemed to cut down on noise from the volume pot. Additionally, with a pot over 10K, C1 must be included as there is DC offset on the pot. What this means is that, unless you want to spend $40 on top quality input caps, 10K is optimal in this design as otherwise there will be a loss in sound quality.

Additionally, I found that different resistors make a big difference here. I built an example with all Holco metal film resistors, and the buffer sounded quite harsh. With all Kiwame carbon film resistors, it is more mellow. With all Kiwame, but with a PRP metal film for R1, it is slightly more detailed than with all Kiwame, but also slightly harder. What this means is that it is worth experimenting to find an optimal sound for your system.

Last, C3 and C4 can be included, especially if the power supply is more than a few inches removed from the circuit. They will tighten up the bass, but with too large a value can cause a tunnel-like sound. I found that 0.22uF Wima caps were a good option in my system, but again it may take some experimenting.

As a last consideration, one needs a power supply. If you don't want to design your own, there are several kits that should work well and that can provide the +/- 15V required. Welborne Labs' PS1 is an excellent choice. You could also use something like Tangent's STEPS or TREAD. With both of those you will need two, or you can build a virtual ground circuit. I built one using a Velleman PS (which is generally pretty low in quaility) but then used dual mono TLE2426's as rail splitters with a 1500uF Panasonic FC cap and a Wima film cap on each rail (that totals to 4 Panasonic caps and 4 Wimas, b/t/w/.) This arrangement is inexpensive and actually quite good, perhaps beter than a single higher quality supply -- the dual mono aspect seperates the channels and helps quite a bit with low level detail. Nuuk also offers some power supply suggestions including using several layers of regulation as well as instructions for a discrete regulator. on the original page. I would think that dual mono is more important than is dual regulation, but certainly dual mono and dual regulation would be the best option. At any rate, the quality of the power will have a direct influence on the quality of the buffer, so it is worth investing some effort here.

While I have only built this design point to point, I have designed a PCB for it. There is space on the board for a Alps pot. Elsewhere I noted that a 10K pot is ideal, though I don't know where to find a 10K Alps (though a web search suggests that they are available). You can get a 50K from either AMB or Tangent. If you can stand to mount it off the board, you can get a 10K Noble from Michael Percy Audio (link is a pdf). I tend to prefer the Noble to the Alps as I feel it has better resolution and theAlps is a bit tinny sounding, but everyone has their opinion here. Last, and highly recommended, you can get a 10K PEC pot from Digikey that, while a bit more expensive, puts the Alps and Noble to shame (people compare it to the highly regarded TKD pots.)

You can download the files to have these boards made from here. (notice that this design is copyrighted and hereby licensed under a creative commons license. It may be used and redistributed for personal, non-commercial, purposes, and you must always credit the original designer. This design is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. Indeed, this design may use lethal voltages. Do not try to build it without sufficient experience to deal with them.) You will need software from ExpressPCB for the board. This company will produce 3 boards for $51.

This design has not been built of protyped. You should do a thorough check for errors on your own as I assume no responsibility for mistakes -- and there most assuredly are a few. If you do decide to have these PCB's made, while you are not required to do so, I would certainly appreciate if you sent me one.